Cláudia Soares

Cláudia Soares

I use optimization, physics, and probability theory to develop robust, interpretable, and trustworthy learning methods to be applied in the real world.

Learning Under Noisy Labels, Spurious Points, and Diverse Structures: TS40K, a 3D Point Cloud Dataset of Rural Terrain and Electrical Transmission Systems

Published in 2025 IEEE/CVF Winter Conference on Applications of Computer Vision (WACV), 2025

Recommended citation: Diogo Lavado, Ricardo Santos, André Coelho, João Santos, Alessandra Micheletti, Claudia Soares, "Learning Under Noisy Labels, Spurious Points, and Diverse Structures: TS40K, a 3D Point Cloud Dataset of Rural Terrain and Electrical Transmission Systems." 2025 IEEE/CVF Winter Conference on Applications of Computer Vision (WACV), 2025. http://dx.doi.org/10.1109/wacv61041.2025.00712

Diogo Lavado, Ricardo Santos, André Coelho, João Santos, Alessandra Micheletti, Claudia Soares

Abstract

Research in 3D scene understanding, particularly in autonomous driving and indoor segmentation, has made significant strides. However, most available datasets focus on urban settings. We introduce TS40K, a 3D point cloud dataset spanning 40,000 km of electrical transmission systems in rural terrain, addressing power-grid inspections to prevent outages, damages, and fires. TS40K offers high point density and no occlusion, presenting challenges like noisy labels, diverse structures, and sensor noise causing spurious points. We evaluate state-of-the-art methods on 3D semantic segmentation and object detection, revealing limitations in power grid inspection. TS40K invites further research to tackle these challenges. Resources available in: https://github.com/dlavado/TS40K

Published abstract source.

Access paper here

BibTeX

@inproceedings{Lavado_2025-2,
    author = "Lavado, Diogo and Santos, Ricardo and Coelho, André and Santos, João and Micheletti, Alessandra and Soares, Claudia",
    title = "Learning Under Noisy Labels, Spurious Points, and Diverse Structures: TS40K, a 3D Point Cloud Dataset of Rural Terrain and Electrical Transmission Systems",
    url = "http://dx.doi.org/10.1109/wacv61041.2025.00712",
    DOI = "10.1109/wacv61041.2025.00712",
    booktitle = "2025 IEEE/CVF Winter Conference on Applications of Computer Vision (WACV)",
    publisher = "IEEE",
    year = "2025",
    month = "Feb",
    pages = "7326-7336",
    abstract = "Research in 3D scene understanding, particularly in autonomous driving and indoor segmentation, has made significant strides. However, most available datasets focus on urban settings. We introduce TS40K, a 3D point cloud dataset spanning 40,000 km of electrical transmission systems in rural terrain, addressing power-grid inspections to prevent outages, damages, and fires. TS40K offers high point density and no occlusion, presenting challenges like noisy labels, diverse structures, and sensor noise causing spurious points. We evaluate state-of-the-art methods on 3D semantic segmentation and object detection, revealing limitations in power grid inspection. TS40K invites further research to tackle these challenges. Resources available in: https://github.com/dlavado/TS40K",
    abstractsource = "https://openaccess.thecvf.com/content/WACV2025/papers/Lavado\_Learning\_under\_Noisy\_Labels\_Spurious\_Points\_and\_Diverse\_Structures\_TS40K\_WACV\_2025\_paper.pdf",
    abstractsourcelabel = "Published abstract source"
}